This project is an extension of previous work involving the creation, development and testing of image processing techniques designed to improve diagnostic performance. Recent work features subtraction of sequentially obtained dental radiographs with the aid of a digital computer. Algorithms which provide bases for quantitative evaluation of the effects of contrast changes and variation in projection geometry are being developed in order to facilitate rejection of method-specific artifacts. Data obtained from double-blind assessment of diagnostic performance show significant improvement over the status quo. Other work involves nonlinear enhancement of CT scans of normal livers containing induced radiolucencies resembling low-contrast, metastatic nodules. By coupling a nonlinear contrast enhancement technique with a simple uniform aperture image convolution, it was possible to achieve a modest increase in diagnostic performance obtained from experienced radiologists.